Behavior-based Risk Detection of Autism Spectrum Disorder Through Child-Robot Interaction

Abstract

This work presents a method to identify children at risk for Autism Spectrum Disorder using behavioral data extracted from video analysis of child-robot interactions. Robots were used as a tool to elicit social engagement from the children in order to capture their social behaviors. A Convolutional Neural Network was used to classify the behavioral data as either at-risk ASD or Typical Development. The network performance was compared to two machine learning classifiers and the utility of the proposed method as a way to streamline existing diagnostic procedures was discussed.